Genomic prediction for numerically small breeds, using models with pre-selected and differentially weighted markers

Table 1 Estimated heritability (h ²_DRP ) obtained by fitting different genomic relationship matrices (GRM) formed from different sets of markers and the corresponding genetic correlation (r_g) between Dutch Holstein (DH) and New Zealand Jersey (NZJ) breeds

GRM fitted	h ²_DRP DH	h ²_DRP NZJ	Estimated r_g (DH vs. NZJ)
MBSG
50k	0.97 (0.00)	0.71 (0.02)	0.22 (0.17)
TOP	0.26 (0.01)	0.27 (0.01)	0.41 (0.15)
COJO8	0.75 (0.01)	0.38 (0.02)	0.44 (0.11)
50k + TOP	0.97 (0.00)	0.71 (0.02)	0.25 (0.17)
50k + COJO8	0.97 (0.00)	0.71 (0.03)	0.29 (0.17)
MBMG
TOP and 50k	0.25 (0.01) and 0.71 (0.01)	0.11 (0.01) and 0.61 (0.02)	0.64 (0.17) and 0.16 (0.19)
COJO8 and 50k	0.26 (0.01) and 0.70 (0.01)	0.17 (0.02) and 0.57 (0.02)	0.88 (0.14) and 0.21 (0.19)

Standard error of estimates are given between parentheses. In the multi-breed, single-GRM model (MBSG), a single multi-breed GRM was fitted in a prediction model, while in the multi-breed, multiple-GRM model (MBMG), two separate multi-breed GRM, formed from different SNP sets, were fitted simultaneously in a prediction model

ISSN: 1297-9686